LEO (computer) - Origins and Initial Design

Origins and Initial Design

J. Lyons and Co., one of the UK's leading catering and food manufacturing companies in the first half of the 20th century, sent two of its senior managers, Oliver Standingford and Raymond Thompson, to the USA in 1947 to look at new business methods developed during the Second World War. During their visit they met Herman Goldstine, one of the original developers of ENIAC, the first general-purpose electronic computer (although it had no stored program). Standingford and Thompson saw the potential of computers to help solve the problem of administering a major business enterprise. They also learned from Goldstine that, back in the UK, Douglas Hartree and Maurice Wilkes were actually building another such machine, the pioneering EDSAC computer, at the University of Cambridge.

On their return to the UK, Standingford and Thompson visited Hartree and Wilkes in Cambridge, and were favourably impressed with their technical expertise and vision. Hartree and Wilkes estimated that EDSAC was twelve to eighteen months from completion, but said that this timeline could be shortened if additional funding were available. Standingford and Thompson wrote a report to the Lyons' Board recommending that Lyons should acquire or build a computer to meet their business needs. The board agreed that, as a first step, Lyons would provide Hartree and Wilkes with £3,000 funding for the EDSAC project, and would also provide them with the services of a Lyons electrical engineer, Ernest Lenaerts. EDSAC was completed and ran its first program in May 1949.

Following the successful completion of EDSAC, the Lyons' board agreed to start the construction of their own machine, expanding on the EDSAC design. The Lyons machine was christened Lyons Electronic Office, or LEO. On the recommendation of Wilkes, Lyons recruited John Pinkerton, a radar engineer and research student at Cambridge, as team leader for the project. Lenaerts returned to Lyons to work on the project, and Wilkes provided training for Lyons' engineer Derek Hemy, who would be responsible for writing LEO's programs. The first business application to be run on LEO was Bakery Valuations. This was initially run as a test program on 5 September 1951, and LEO took over Bakery Valuations calculations completely on 29 November 1951.